Abstract
Multiple sclerosis is a chronic, immune-mediated neurological disorder affecting more than 2.8 million people worldwide. In approximately 85% of cases, the disease first presents as clinically isolated syndrome, an acute neurological episode with a variable risk of progressing to multiple sclerosis. Early identification of high-risk patients is critical for timely intervention. This study analyzes a cohort of 273 Mexican mestizo patients diagnosed with clinically isolated syndrome at the National Institute of Neurology and Neurosurgery in Mexico City between 2006 and 2010. We applied and compared multiple machine learning models including logistic regression, k-nearest neighbors, naïve Bayes, support vector machine, and random forest using demographic, clinical, and magnetic resonance imaging features to predict progression to multiple sclerosis. Among these models, the support vector machine with a radial basis function kernel achieved the highest accuracy (80.9%), while logistic regression provided interpretable insights into key predictors. Across models, magnetic resonance imaging findings particularly periventricular and spinal cord lesions emerged as the most influential predictors, with age also contributing to risk stratification. These results underscore the potential of machine learning to support early diagnosis, guide personalized treatment strategies, and improve outcomes for patients at high risk of developing multiple sclerosis.